I
AE
S
I
nte
rna
t
io
na
l J
o
urna
l o
f
Ro
bo
t
ics a
nd
Aut
o
m
a
t
io
n
(
I
J
RA)
Vo
l.
1
0
,
No
.
4
,
Dec
em
b
er
2
0
2
1
,
p
p
.
3
2
6
~
3
3
9
I
SS
N:
2722
-
2
5
8
6
,
DOI
: 1
0
.
1
1
5
9
1
/
i
jr
a
.
v
1
0
i
4
.
pp
3
2
6
-
339
326
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
r
a
.
ia
esco
r
e.
co
m
Ada
ptive la
ng
ua
g
e process
ing
uni
t
for Ma
la
y
sia
n sig
n lang
ua
g
e
sy
nthesizer
H
a
ris Al
Q
o
dri Ma
a
rif
1
,
T
eddy
Su
ry
a
G
un
a
wa
n
2
,
Rini A
k
m
elia
wa
t
i
3
1
De
p
a
rtme
n
t
o
f
M
e
c
h
a
tr
o
n
ics
E
n
g
in
e
e
rin
g
,
I
n
tern
a
ti
o
n
a
l
Isla
m
ic U
n
iv
e
rsit
y
M
a
lay
sia
,
M
a
lay
sia
2
De
p
a
rtme
n
t
o
f
El
e
c
tri
c
a
l
a
n
d
Co
m
p
u
ter E
n
g
i
n
e
e
rin
g
,
In
ter
n
a
ti
o
n
a
l
Isla
m
ic Un
iv
e
rsity
M
a
lay
sia
,
M
a
lay
sia
2
S
c
h
o
o
l
o
f
El
e
c
tri
c
a
l
En
g
in
e
e
ri
n
g
a
n
d
Tele
c
o
m
m
u
n
ica
ti
o
n
s,
Un
i
v
e
rsity
o
f
Ne
w S
o
u
th
Wale
s,
Au
stra
l
ia
3
S
c
h
o
o
l
o
f
M
e
c
h
a
n
ica
l
E
n
g
i
n
e
e
ri
n
g
,
U
n
iv
e
rsit
y
o
f
Ad
e
laid
e
,
Au
str
a
li
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
2
4
J
u
n
,
2
0
2
1
R
ev
is
ed
29
J
u
l
,
2
0
2
1
Acc
ep
ted
2
6
Au
g
,
2
0
2
1
Lan
g
u
a
g
e
p
r
o
c
e
ss
in
g
u
n
it
(LP
U)
is
a
sy
ste
m
b
u
il
t
to
p
r
o
c
e
ss
tex
t
-
b
a
se
d
d
a
ta
to
c
o
m
p
ly
wit
h
t
h
e
r
u
les
o
f
th
e
sig
n
lan
g
u
a
g
e
g
ra
m
m
a
r.
Th
is
s
y
ste
m
wa
s
d
e
v
e
lo
p
e
d
a
s
a
n
imp
o
rtan
t
p
a
rt
o
f
th
e
sig
n
lan
g
u
a
g
e
sy
n
th
e
siz
e
r
sy
ste
m
.
S
ig
n
lan
g
u
a
g
e
(S
L)
u
se
s
d
iffere
n
t
g
ra
m
m
a
ti
c
a
l
ru
les
fro
m
t
h
e
sp
o
k
e
n
/v
e
r
b
a
l
lan
g
u
a
g
e
,
wh
ich
o
n
l
y
in
v
o
lv
e
s
t
h
e
imp
o
rtan
t
wo
rd
s
t
h
a
t
h
e
a
rin
g
/i
m
p
a
ired
sp
e
e
c
h
p
e
o
p
le
c
a
n
u
n
d
e
rsta
n
d
.
Th
e
re
fo
re
,
it
n
e
e
d
s
wo
r
d
c
las
sifica
ti
o
n
b
y
LP
U
to
d
e
term
in
e
g
ra
m
m
a
ti
c
a
ll
y
p
ro
c
e
ss
e
d
se
n
ten
c
e
s
fo
r
th
e
sig
n
lan
g
u
a
g
e
sy
n
th
e
siz
e
r.
H
o
we
v
e
r,
th
e
e
x
i
stin
g
lan
g
u
a
g
e
p
r
o
c
e
ss
in
g
u
n
i
t
in
S
L
sy
n
th
e
siz
e
rs
su
ffe
rs
ti
m
e
lag
g
in
g
a
n
d
c
o
m
p
le
x
it
y
p
r
o
b
lem
s,
re
su
lt
i
n
g
i
n
h
ig
h
p
ro
c
e
ss
in
g
ti
m
e
.
T
h
e
two
fe
a
tu
re
s,
i.
e
.
,
th
e
c
o
m
p
u
tati
o
n
a
l
ti
m
e
a
n
d
su
c
c
e
ss
ra
te,
b
e
c
o
m
e
trad
e
-
o
ffs
w
h
ich
m
e
a
n
s
t
h
e
p
r
o
c
e
ss
in
g
ti
m
e
b
e
c
o
m
e
s
lo
n
g
e
r
to
a
c
h
iev
e
a
h
ig
h
e
r
su
c
c
e
ss
ra
te.
Th
is
p
a
p
e
r
p
r
o
p
o
se
s
a
n
a
d
a
p
ti
v
e
LP
U
th
a
t
a
ll
o
ws
p
ro
c
e
ss
in
g
t
h
e
wo
rd
s
fro
m
sp
o
k
e
n
wo
rd
s
t
o
M
a
l
a
y
sia
n
S
L
g
ra
m
m
a
ti
c
a
l
ru
le
th
a
t
re
su
l
ts
i
n
re
lativ
e
ly
fa
st
p
ro
c
e
ss
in
g
ti
m
e
a
n
d
a
g
o
o
d
su
c
c
e
ss
r
a
te.
It
in
v
o
l
v
e
s
n
-
g
ra
m
s,
n
a
tu
ra
l
lan
g
u
a
g
e
p
r
o
c
e
ss
in
g
(
NLP
)
,
a
n
d
h
id
d
e
n
M
a
r
k
o
v
m
o
d
e
ls
(HMM
)/Bay
e
sia
n
n
e
two
r
k
s
a
s
th
e
c
l
a
ss
ifi
e
r
to
p
ro
c
e
ss
th
e
tex
t
-
b
a
se
d
i
n
p
u
t.
A
s
a
re
su
lt
,
th
e
p
ro
p
o
se
d
LP
U
s
y
ste
m
h
a
s
su
c
c
e
ss
fu
ll
y
p
r
o
v
i
d
e
d
a
n
e
fficie
n
t
(fa
st)
p
ro
c
e
ss
in
g
t
ime
a
n
d
a
g
o
o
d
su
c
c
e
ss
ra
te
c
o
m
p
a
re
d
to
LP
U
wit
h
o
t
h
e
r
e
d
it
d
istan
c
e
s
(
m
a
h
a
lan
o
b
is
,
Le
v
e
n
sh
tei
n
,
a
n
d
so
u
n
d
e
x
).
Th
e
sy
ste
m
h
a
s
b
e
e
n
tes
ted
o
n
1
3
0
tex
t
-
i
n
p
u
t
se
n
t
e
n
c
e
s
with
se
v
e
ra
l
wo
rd
s
ra
n
g
in
g
fr
o
m
3
t
o
1
0
w
o
rd
s.
Re
su
lt
s
s
h
o
we
d
th
a
t
th
e
p
ro
p
o
se
d
LP
U
c
o
u
ld
a
c
h
iev
e
a
ro
u
n
d
1
.
4
9
7
m
s
p
ro
c
e
ss
in
g
ti
m
e
wit
h
a
n
a
v
e
ra
g
e
su
c
c
e
ss
ra
te o
f
8
4
.
2
3
%
fo
r
a
m
a
x
im
u
m
o
f
ten
-
wo
rd
se
n
ten
c
e
s.
K
ey
w
o
r
d
s
:
C
las
s
if
ier
Dis
tan
ce
alg
o
r
ith
m
Ma
lay
s
ian
s
ig
n
lan
g
au
g
e
Natu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Har
is
Al
Qo
d
r
i M
aa
r
if
Dep
ar
tm
en
t o
f
Me
ch
atr
o
n
ics E
n
g
in
ee
r
in
g
I
n
ter
n
atio
n
al
I
s
lam
ic
Un
iv
er
s
ity
Ma
lay
s
ia
J
alan
Go
m
b
ak
,
Ku
ala
L
u
m
p
u
r
,
5
3
1
0
0
,
Ma
lay
s
ia
E
m
ail: a
lq
o
d
r
i.m
aa
r
if
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Sig
n
lan
g
u
ag
e
(
SL)
is
th
e
p
r
i
m
ar
y
lan
g
u
ag
e
an
d
ca
n
b
e
c
o
n
s
id
er
ed
as
th
e
m
o
th
er
to
n
g
u
e
f
o
r
th
e
HSI
p
eo
p
le.
Ma
n
y
p
e
o
p
le
wh
o
ar
e
b
o
r
n
d
ea
f
lear
n
s
ig
n
lan
g
u
ag
e
as
th
eir
p
r
im
ar
y
lan
g
u
ag
e,
a
n
d
it
r
em
ain
s
th
eir
p
r
ef
er
r
e
d
,
o
r
f
ir
s
t,
lan
g
u
ag
e
.
T
h
er
e
is
n
o
wr
itten
f
o
r
m
o
f
s
ig
n
la
n
g
u
ag
e,
s
o
d
ea
f
p
eo
p
le
co
m
m
u
n
icate
u
s
in
g
r
ea
d
in
g
a
n
d
wr
itin
g
i
n
th
eir
s
ec
o
n
d
o
r
less
p
r
ef
er
r
e
d
lan
g
u
ag
e.
T
h
er
e
f
o
r
e,
a
s
ig
n
if
ican
t
p
r
o
p
o
r
tio
n
o
f
d
ea
f
p
eo
p
le
h
av
e
a
s
tr
o
n
g
p
r
ef
er
e
n
ce
f
o
r
ac
ce
s
s
in
g
in
f
o
r
m
atio
n
in
s
ig
n
lan
g
u
ag
e
r
ath
e
r
th
an
wr
itten
tex
t.
Sig
n
l
an
g
u
ag
e
o
n
ly
n
ee
d
s
s
o
m
e
im
p
o
r
tan
t
wo
r
d
s
co
m
p
ar
e
d
to
s
p
o
k
en
lan
g
u
ag
e
[
1
]
,
[
2
]
.
Gen
e
r
ally
,
s
ig
n
lan
g
u
ag
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t
J
R
o
b
&
A
u
to
m
I
SS
N:
2722
-
2
5
8
6
A
d
a
p
tive
la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
u
n
it fo
r
Ma
la
ysia
n
s
ig
n
la
n
g
u
a
g
e
s
yn
th
esiz
er
(
Ha
r
is
A
l Q
o
d
r
i Ma
a
r
if
)
327
u
s
es
s
u
b
ject,
v
er
b
,
n
o
u
n
,
an
d
ad
v
er
b
.
T
h
e
r
e
ar
e
n
o
o
th
er
s
u
f
f
ix
es,
p
r
e
f
ix
es,
an
d
p
ar
ticles.
I
t
is
a
n
o
n
-
v
er
b
al
lan
g
u
ag
e
th
at
u
s
es
h
an
d
m
o
v
em
en
t,
h
an
d
o
r
ien
tatio
n
,
f
ac
e
ex
p
r
ess
io
n
,
h
ea
d
m
o
v
e
m
en
t,
p
o
s
tu
r
e,
an
d
b
o
d
y
o
r
ien
tatio
n
[
3
]
.
Sin
ce
s
ig
n
lan
g
u
ag
e
is
a
n
o
n
-
v
er
b
al
la
n
g
u
a
g
e,
th
e
u
n
d
er
s
tan
d
in
g
o
f
s
ig
n
lan
g
u
ag
e
h
as
b
ee
n
co
m
p
u
ls
o
r
y
f
o
r
HSI
p
eo
p
le
to
co
m
m
u
n
icate
.
T
h
e
awa
r
en
ess
o
f
s
ig
n
lan
g
u
a
g
e
f
o
r
n
o
n
-
HSI
p
eo
p
le
is
litt
le,
o
r
m
an
y
d
o
n
o
t
k
n
o
w
s
ig
n
lan
g
u
a
g
e.
T
h
er
ef
o
r
e,
it
p
r
o
v
id
es
an
o
b
s
tac
le
in
co
m
m
u
n
icatio
n
in
th
e
co
m
m
u
n
ity
,
esp
ec
ially
if
it
n
ee
d
s
in
ter
ac
tio
n
b
etwe
en
n
o
n
-
HSI
a
n
d
HSI
p
eo
p
le.
As
o
b
s
tacle
s
ar
is
e
in
co
n
tact
with
t
h
e
co
m
m
u
n
ity
,
th
e
c
o
m
m
u
n
icatio
n
b
r
id
g
e
m
u
s
t f
ill th
e
g
ap
b
etwe
en
th
em
.
T
h
e
o
p
tio
n
s
ar
e
s
ig
n
lan
g
u
ag
e
tr
a
n
s
lato
r
an
d
s
ig
n
la
n
g
u
ag
e
s
y
n
th
esizer
tech
n
o
lo
g
y
,
tr
an
s
latin
g
s
p
o
k
en
lan
g
u
ag
e
to
s
ig
n
lan
g
u
ag
e
[
4
]
.
Usi
n
g
a
s
ig
n
lan
g
u
ag
e
tr
an
s
lato
r
to
co
m
m
u
n
icate
b
etwe
en
n
o
n
-
HSI
an
d
HSI
h
as
b
ee
n
li
m
ited
s
in
ce
s
ig
n
lan
g
u
a
g
e
tr
an
s
la
to
r
s
ar
e
l
im
ited
in
Ma
lay
s
ia.
As
ea
r
ly
as
2
0
1
7
,
th
er
e
a
r
e
o
n
ly
less
th
an
1
0
0
ce
r
tifie
d
SL
tr
an
s
lato
r
s
to
ca
ter
to
m
o
r
e
t
h
an
3
0
,
0
0
0
p
e
r
s
o
n
s
o
f
HSI
(
r
ef
er
en
ce
s
)
.
W
h
ile
in
th
e
wo
r
ld
,
t
h
e
wo
r
ld
f
ed
e
r
atio
n
o
f
th
e
d
e
af
r
ep
o
r
ted
th
at
th
er
e
ar
e
ab
o
u
t
7
0
m
illi
o
n
HS
I
p
eo
p
le
[
4
]
an
d
1
3
8
liv
in
g
s
ig
n
lan
g
u
ag
e,
wh
ich
is
ac
co
r
d
in
g
to
th
e
e
th
n
o
lo
g
u
e
ca
talo
g
[
5
]
.
Sig
n
lan
g
u
ag
e
s
y
n
th
esizer
c
o
n
s
is
ts
o
f
th
r
ee
m
ain
m
o
d
u
l
es,
i.e
.
,
t
h
e
v
o
ice
r
ec
o
g
n
itio
n
m
o
d
u
le,
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
u
n
it
m
o
d
u
le,
an
d
s
ig
n
in
g
m
o
d
u
le.
E
ac
h
m
o
d
u
le
h
as
its
co
m
p
o
n
e
n
ts
an
d
alg
o
r
ith
m
s
wh
ich
n
ee
d
a
d
if
f
er
en
t
ap
p
r
o
a
ch
to
d
ev
elo
p
m
e
n
t.
I
n
th
is
p
ap
er
,
th
e
m
ain
f
o
cu
s
is
o
n
th
e
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
m
o
d
u
le,
wh
ich
tr
a
n
s
f
o
r
m
s
th
e
in
p
u
t
lan
g
u
a
g
e.
T
h
e
lan
g
u
a
g
e
p
r
o
ce
s
s
in
g
m
o
d
u
le
alter
s
in
p
u
t
lan
g
u
ag
e
in
to
o
u
tp
u
t
lan
g
u
ag
e
th
at
is
s
u
ita
b
le
f
o
r
o
u
tp
u
t
s
ig
n
lan
g
u
ag
e.
T
h
e
in
p
u
t
an
d
o
u
tp
u
t
lan
g
u
ag
e
ar
e
in
th
e
s
eq
u
en
ce
o
f
wo
r
d
s
(
te
x
t)
,
in
w
h
ich
s
o
m
e
m
eth
o
d
o
lo
g
y
is
r
eq
u
ir
ed
to
d
o
th
e
tr
an
s
f
o
r
m
atio
n
p
r
o
ce
s
s
p
r
o
p
er
ly
.
T
h
e
d
ev
el
o
p
m
e
n
t
o
f
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
u
n
its
h
as
b
ee
n
m
ad
e
an
d
im
p
lem
e
n
ted
in
m
an
y
d
if
f
e
r
en
t
s
ig
n
lan
g
u
ag
es,
f
o
r
ex
am
p
le,
Am
er
ican
s
ig
n
lan
g
u
ag
e
[
6
]
,
B
r
itis
h
s
ig
n
lan
g
u
ag
e
[
7
]
,
So
u
th
Af
r
ica
n
s
ig
n
lan
g
u
ag
e
[
8
]
,
an
d
Au
s
tr
alian
s
ig
n
lan
g
u
ag
e
[
9
]
.
Ho
wev
er
,
i
n
Ma
lay
s
ia,
th
e
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
u
n
it
h
as
n
o
t
b
ee
n
im
p
lem
e
n
ted
as
an
in
teg
r
al
p
ar
t
o
f
th
e
s
ig
n
lan
g
u
ag
e
s
y
n
th
esizer
.
Fu
r
th
er
m
o
r
e,
th
e
l
an
g
u
ag
e
p
r
o
ce
s
s
in
g
u
n
it
f
o
r
b
a
h
asa
is
y
ar
at
Ma
lay
s
ia
h
as
n
o
t
b
ee
n
im
p
lem
en
ted
.
A
co
m
p
r
eh
en
s
iv
e
r
e
v
iew
o
f
th
e
ex
is
tin
g
wo
r
k
an
d
p
r
o
p
o
s
ed
wo
r
k
o
n
th
e
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
u
n
it
is
p
r
esen
ted
in
th
is
p
ap
er
.
I
n
ad
d
itio
n
,
v
ar
io
u
s
m
eth
o
d
s
s
u
ch
as
ed
it d
is
tan
ce
,
n
atu
r
al
l
an
g
u
ag
e
p
r
o
ce
s
s
in
g
,
HM
M
m
eth
o
d
s
,
an
d
B
ay
esian
n
etwo
r
k
ar
e
d
is
cu
s
s
ed
.
2.
L
I
T
E
R
AT
U
RE
R
E
VI
E
W
T
h
is
s
ec
tio
n
r
ev
iews
th
e
tech
n
iq
u
es
o
f
th
e
lan
g
u
a
g
e
p
r
o
ce
s
s
in
g
u
n
it.
T
h
e
tec
h
n
iq
u
e
f
o
r
th
e
lan
g
u
a
g
e
p
r
o
ce
s
s
in
g
u
n
it
p
r
o
v
id
es
a
lit
er
atu
r
e
b
ac
k
g
r
o
u
n
d
f
o
r
th
e
l
an
g
u
ag
e
p
r
o
ce
s
s
in
g
u
n
it
u
s
in
g
n
atu
r
al
lan
g
u
a
g
e
p
r
o
ce
s
s
in
g
.
2
.
1
.
Na
t
ura
l
la
ng
ua
g
e
pro
ce
s
s
ing
(
NL
P
)
2
.
1
.
1
.
NL
P
ba
s
ic
pro
ce
s
s
ing
T
h
e
n
ec
ess
ar
y
p
r
o
ce
s
s
o
f
Nat
u
r
al
lan
g
u
a
g
e
p
r
o
ce
s
s
in
g
(
NL
P
)
th
at
ca
n
b
e
u
s
ed
f
o
r
SL
s
y
n
th
esizer
is
th
e
m
o
s
t
s
tr
aig
h
tf
o
r
war
d
tec
h
n
iq
u
e
wh
ic
h
h
as
b
ee
n
im
p
lem
en
ted
[
10
]
.
T
h
is
tech
n
i
q
u
e
o
n
ly
in
v
o
lv
es
th
r
ee
b
asic
o
p
er
atio
n
s
,
i.e
.
,
POS
(
p
a
r
t
o
f
s
p
ee
ch
)
tag
g
er
,
o
p
tim
ize
r
,
an
d
s
tem
m
in
g
.
Fig
u
r
e
1
s
h
o
ws
th
e
s
tep
o
f
th
is
tech
n
iq
u
e.
Fig
u
r
e
1
.
NL
P
b
asic p
r
o
ce
s
s
in
g
[
4
]
T
h
e
in
itial
s
tag
e
i
s
th
e
PO
S
tag
g
er
,
wh
ich
in
v
o
lv
es
m
o
r
p
h
o
lo
g
ical
an
aly
s
is
.
T
h
en
,
as
th
e
PO
S
tag
g
er
'
s
o
u
tp
u
t,
th
e
o
p
tim
izer
tak
es
p
ar
t
in
th
e
f
o
llo
win
g
s
tep
to
r
em
o
v
e
th
e
u
n
n
ec
ess
ar
y
wo
r
d
s
.
Fin
ally
,
b
ef
o
r
e
t
h
e
o
u
t
p
u
t
is
g
iv
e
n
to
t
h
e
an
im
atio
n
s
tep
[
11
]
,
th
e
s
t
em
m
in
g
p
r
o
ce
s
s
is
in
v
o
lv
ed
in
f
in
d
in
g
th
e
w
o
r
d
s
'
p
r
im
ar
y
f
o
r
m
.
2
.
1
.
2
.
NL
P
wit
h g
lo
s
s
-
ba
s
e
d
a
pp
ro
a
ch
T
h
e
g
lo
s
s
-
b
as
ed
ap
p
r
o
ac
h
is
a
m
eth
o
d
th
at
ass
o
ciate
s
th
e
wo
r
d
s
an
d
th
eir
m
ea
n
in
g
s
th
r
o
u
g
h
a
d
ictio
n
ar
y
[
12
]
.
T
h
e
o
r
d
er
o
f
t
h
e
lan
g
u
ag
e
g
r
am
m
a
r
d
ef
in
es
th
e
o
r
d
er
o
f
t
h
e
g
l
o
s
s
es.
I
n
a
r
ep
o
r
t
b
y
Alm
eid
a
et
a
l
.
[
1
3
]
,
th
e
o
r
d
er
o
f
b
lo
ck
s
(
g
lo
s
s
es)
is
ca
lcu
lated
ac
co
r
d
in
g
to
Po
r
tu
g
al
s
ig
n
lan
g
u
ag
e
(
L
GP)
g
r
am
m
ar
.
As
th
e
f
in
al
s
tep
,
th
e
o
r
d
er
o
f
b
lo
ck
s
was
co
n
v
e
r
ted
in
to
t
h
e
s
ig
n
lan
g
u
ag
e
o
r
d
er
.
Fig
u
r
e
2
s
h
o
ws
th
e
g
l
o
s
s
-
b
ased
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2722
-
2
5
8
6
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
,
Vo
l
.
1
0
,
No
.
4
,
Dec
em
b
er
2
0
2
1
:
3
2
6
–
339
328
ap
p
r
o
ac
h
tech
n
i
q
u
e
b
y
Alm
e
id
a
et
a
l
.
[
1
3
]
.
T
h
e
tex
t
was
ass
o
ciate
d
with
th
e
d
ictio
n
a
r
y
(
d
ata
b
ase)
to
b
e
p
r
o
ce
s
s
ed
in
a
n
atu
r
al
lan
g
u
a
g
e
p
r
o
ce
s
s
in
g
b
lo
c
k
.
I
n
a
d
d
it
io
n
,
th
e
o
u
t
p
u
t
was
tr
an
s
lated
to
a
s
eq
u
en
ce
o
f
g
lo
s
s
es a
n
d
ac
tio
n
s
.
A
s
tem
m
er
ca
n
b
e
u
s
ed
t
o
i
d
en
tify
th
e
s
tem
an
d
r
elev
an
t
s
u
f
f
ix
es
(
an
d
p
r
efix
es),
wh
i
ch
allo
ws
in
f
er
r
in
g
,
f
o
r
in
s
tan
ce
,
th
e
g
e
n
d
er
an
d
th
e
n
u
m
b
er
o
f
g
i
v
e
n
wo
r
d
s
.
A
p
ar
t
-
of
-
s
p
ee
ch
(
P
OS)
tag
g
er
ca
n
also
co
n
tr
ib
u
te
t
o
th
e
t
r
an
s
latio
n
p
r
o
ce
s
s
,
wh
ich
c
o
u
p
les
with
th
e
s
tem
m
er
in
t
h
e
id
en
tifi
ca
t
io
n
o
f
th
e
d
if
f
er
e
n
t
ty
p
es
o
f
af
f
ix
es.
I
n
ad
d
itio
n
,
a
POS
tag
g
er
u
s
u
ally
f
ee
d
s
f
u
r
th
er
p
r
o
ce
s
s
in
g
,
f
o
r
in
s
ta
n
ce
,
n
am
ed
en
tity
r
ec
o
g
n
izer
s
an
d
s
y
n
tactic
an
aly
ze
r
s
.
A
n
am
ed
e
n
tity
r
ec
o
g
n
izer
allo
ws
id
en
tify
in
g
p
e
r
s
o
n
s
'
n
am
es
an
d
a
s
y
n
tactic
an
aly
ze
r
to
d
eter
m
in
e
th
e
s
en
ten
ce
'
s
s
y
n
tactic
co
m
p
o
n
en
ts
,
s
u
ch
as su
b
ject
a
n
d
o
b
ject.
Fig
u
r
e
2
.
NL
P with
g
lo
s
s
b
ase
d
tech
n
iq
u
e
[
13
]
2
.
1
.
3
.
NL
P
wit
h
rule
-
ba
s
e
d a
nd
s
t
a
t
is
t
ica
l t
ra
ns
la
t
io
n
T
h
e
r
u
le
-
b
ased
tr
an
s
latio
n
is
a
s
tr
ateg
y
th
at
an
aly
s
es
th
e
wo
r
d
'
s
in
p
u
t
u
n
til
a
g
r
o
u
p
o
f
wo
r
d
s
(
s
en
ten
ce
)
[
14
]
.
T
h
e
tr
a
n
s
latio
n
an
al
y
s
is
f
in
d
s
s
p
ec
if
ic
co
m
b
in
atio
n
s
o
f
wo
r
d
s
o
r
s
ig
n
s
(
b
l
o
ck
s
)
th
at
g
en
e
r
ate
a
s
ig
n
.
T
h
e
f
i
n
d
in
g
p
r
o
ce
s
s
s
tar
ts
f
r
o
m
ea
ch
w
o
r
d
in
d
iv
id
u
ally
an
d
e
x
ten
d
s
th
e
a
n
aly
s
is
to
n
eig
h
b
o
r
h
o
o
d
co
n
tex
t
w
o
r
d
s
o
r
alr
ea
d
y
-
f
o
r
m
ed
s
ig
n
s
.
T
h
er
e
ar
e
two
s
te
p
s
in
v
o
lv
ed
in
t
h
e
tr
an
s
latio
n
p
r
o
ce
s
s
.
I
n
th
e
f
ir
s
t
o
n
e,
ev
e
r
y
w
o
r
d
is
m
a
p
p
ed
to
o
n
e
o
r
s
ev
e
r
al
s
y
n
tactic
p
r
a
g
m
atic
tag
s
.
T
h
e
t
r
an
s
latio
n
m
o
d
u
le
t
h
en
ap
p
lies
d
if
f
er
en
t
r
u
les
th
at
co
n
v
e
r
t
th
e
tag
g
ed
w
o
r
d
s
in
to
s
ig
n
s
t
h
r
o
u
g
h
g
r
o
u
p
in
g
c
o
n
ce
p
ts
o
r
s
ig
n
s
(
b
lo
ck
s
)
an
d
d
ef
in
in
g
n
ew
s
ig
n
s
.
T
h
ese
r
u
l
es
ca
n
d
ef
in
e
s
h
o
r
t
an
d
ex
ten
s
iv
e
s
co
p
e
r
elatio
n
s
h
ip
s
b
etw
ee
n
th
e
co
n
ce
p
ts
o
r
s
ig
n
s
.
At
th
e
e
n
d
o
f
th
e
p
r
o
ce
s
s
,
th
e
b
lo
ck
s
eq
u
en
ce
is
ex
p
e
cted
to
c
o
r
r
esp
o
n
d
to
t
h
e
s
ig
n
s
eq
u
en
ce
r
esu
ltin
g
f
r
o
m
th
e
tr
an
s
latio
n
p
r
o
ce
s
s
.
T
h
e
r
u
le
-
b
ased
tr
a
n
s
latio
n
m
o
d
u
le
p
r
o
v
id
es
th
e
tr
an
s
l
atio
n
r
u
les
f
o
r
th
e
tr
an
s
latio
n
p
r
o
ce
s
s
.
San
-
Seg
u
n
d
o
et
a
l
.
[
1
5
]
p
r
o
v
id
ed
th
e
e
v
alu
atio
n
to
o
ls
f
o
r
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
es.
T
h
r
ee
av
ailab
le
m
ea
s
u
r
em
e
n
t
to
o
ls
h
av
e
b
ee
n
c
o
n
s
id
er
ed
:
s
ig
n
er
r
o
r
r
ate
(
SER),
p
o
s
itio
n
in
d
e
p
en
d
en
t
r
ate
(
PER),
an
d
b
ilin
g
u
al
ev
alu
atio
n
u
n
d
er
s
tu
d
y
(
B
L
E
U)
.
T
h
e
s
tatis
tical
tr
an
s
latio
n
m
eth
o
d
ca
lcu
la
tes
th
e
p
r
o
b
ab
ilit
y
b
etwe
en
th
e
w
o
r
d
s
eq
u
en
ce
an
d
s
ig
n
s
eq
u
en
ce
s
to
r
ed
in
a
d
atab
ase
as
th
e
r
ef
er
en
ce
[
16
]
.
On
e
o
f
th
e
m
eth
o
d
s
in
s
tatis
tical
tr
an
s
latio
n
is
p
h
r
ase
-
b
ased
tr
an
s
latio
n
[
15
]
,
[
17
]
.
Fig
u
r
e
3
s
h
o
ws
th
e
d
iag
r
a
m
o
f
t
h
e
p
h
r
ase
-
b
ased
tr
a
n
s
latio
n
m
o
d
u
le
u
s
ed
b
y
San
-
Seg
u
n
d
o
et
a
l
.
[
1
5
]
.
Fig
u
r
e
3
.
Diag
r
a
m
o
f
p
h
r
ase
-
b
ased
tr
an
s
latio
n
m
o
d
u
le
[
18
]
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t
J
R
o
b
&
A
u
to
m
I
SS
N:
2722
-
2
5
8
6
A
d
a
p
tive
la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
u
n
it fo
r
Ma
la
ysia
n
s
ig
n
la
n
g
u
a
g
e
s
yn
th
esiz
er
(
Ha
r
is
A
l Q
o
d
r
i Ma
a
r
if
)
329
T
h
e
tr
an
s
latio
n
p
r
o
ce
s
s
u
s
es
a
tr
an
s
latio
n
m
o
d
el
b
ased
o
n
p
h
r
ases
an
d
a
ta
r
g
et
la
n
g
u
a
g
e
m
o
d
el
as
r
ep
o
r
ted
i
n
San
-
Seg
u
n
d
o
et
a
l
.
[
1
9
]
,
[
2
0
]
,
th
e
GI
Z
A+
+
s
o
f
twar
e
h
as
b
ee
n
u
s
ed
to
ca
lc
u
late
th
e
alig
n
m
en
ts
b
etwe
en
wo
r
d
s
an
d
s
ig
n
s
.
Sa
n
-
Seg
u
n
d
o
et
a
l
.
[
1
5
]
r
e
p
o
r
te
d
th
at
t
h
e
s
tatis
tical
tr
an
s
latio
n
s
h
o
ws
th
e
wo
r
s
t
o
u
tco
m
e
f
r
o
m
th
e
r
u
le
-
b
ased
s
tr
ateg
y
.
T
h
is
co
n
d
itio
n
is
d
u
e
to
its
r
estricte
d
d
o
m
ain
,
an
d
i
t
h
as
b
ee
n
p
o
s
s
ib
le
to
d
ev
elo
p
a
co
m
p
lete
s
et
o
f
r
u
les with
a
r
ea
s
o
n
ab
le
ef
f
o
r
t.
2
.
2
.
E
dit
di
s
t
a
nce
t
ec
hn
iqu
e
s
2
.
2
.
1
.
L
ev
ens
hte
i
n
dis
t
a
nce
L
ev
en
s
h
tein
d
is
tan
ce
(
L
D)
is
a
tech
n
i
q
u
e
f
o
r
lo
o
k
in
g
f
o
r
th
e
d
i
f
f
er
en
ce
s
b
etwe
en
tw
o
d
if
f
e
r
en
t
s
tr
in
g
s
an
d
co
m
p
u
tin
g
th
e
two
d
if
f
er
en
t
p
h
o
n
etic
s
tr
in
g
s
'
d
i
s
tan
ce
.
T
h
e
b
asic
tech
n
iq
u
e
o
f
L
D
in
v
o
lv
es
th
r
ee
m
ain
p
r
o
ce
s
s
es,
wh
ich
ar
e
in
s
er
tio
n
,
d
eletio
n
,
an
d
s
u
b
s
titu
tio
n
s
.
T
h
e
L
D
is
a
m
eth
o
d
o
f
al
ig
n
in
g
two
p
h
o
n
etic
s
eg
m
en
ts
.
T
h
e
en
h
a
n
ce
m
en
t
was
im
p
lem
en
ted
i
n
p
r
io
r
r
ese
ar
ch
,
allo
win
g
o
n
ly
alig
n
m
e
n
ts
o
f
c
o
n
s
o
n
a
n
ts
with
co
n
s
o
n
an
ts
an
d
v
o
wels with
v
o
wels
[
21
]
.
T
h
e
L
D
is
a
p
o
p
u
lar
s
tr
in
g
m
etr
ic
u
s
ed
to
ev
alu
a
te
s
tr
in
g
s
o
n
o
r
th
o
g
r
ap
h
ic
s
im
ilar
ity
in
i
n
f
o
r
m
atio
n
th
eo
r
y
.
L
D
c
o
u
n
ts
m
in
im
al
s
u
b
s
titu
tio
n
s
,
in
s
er
tio
n
s
,
an
d
d
el
etio
n
s
to
ed
it o
n
e
s
tr
in
g
in
to
a
n
o
th
er
o
f
an
y
le
n
g
th
[
22
]
.
Fo
r
wo
r
d
p
air
s
with
eq
u
al
wo
r
d
len
g
th
,
L
D
p
r
o
d
u
ce
s
o
n
ly
d
is
tan
ce
s
s
m
aller
o
r
eq
u
al
to
th
e
Ham
m
in
g
d
is
tan
ce
[
23
]
.
T
h
e
Ham
m
in
g
d
is
tan
ce
co
u
n
ts
th
e
m
in
im
al
n
u
m
b
er
o
f
s
u
b
s
titu
tio
n
s
n
ee
d
ed
to
ed
it
o
n
e
s
tr
in
g
in
to
an
o
th
e
r
eq
u
al
le
n
g
th
[
24
]
.
T
h
e
im
p
lem
e
n
tatio
n
o
f
L
D
p
r
o
v
id
es
th
e
d
is
ta
n
ce
ca
lc
u
latio
n
o
f
two
v
ar
ieties
o
f
wo
r
d
s
[
25
]
.
T
h
e
L
D
was
u
s
ed
in
th
e
an
aly
s
is
o
f
li
n
g
u
is
tic
v
ar
iat
io
n
s
in
m
a
n
y
o
th
er
lan
g
u
ag
es,
f
o
r
ex
am
p
le,
Ger
m
an
[
26
]
,
Du
tc
h
[
27
]
,
Frisian
[
28
]
,
a
n
d
B
u
lg
a
r
ian
[
29
]
.
T
h
e
o
th
er
s
u
cc
ess
f
u
l
im
p
lem
en
tatio
n
was
test
ed
o
n
1
5
No
r
weg
ian
d
ialec
ts
p
er
ce
p
tu
ally
an
d
ac
o
u
s
tically
[
30
]
.
2
.
2
.
2
.
M
a
ha
la
no
bis
di
s
t
a
nce
Ma
h
alan
o
b
is
d
is
tan
ce
is
th
e
d
is
tan
ce
b
etwe
en
two
s
am
p
le
s
b
ased
o
n
th
eir
m
ea
n
f
ea
t
u
r
e
v
ec
to
r
s
an
d
,
an
d
th
e
co
v
ar
ian
ce
m
at
r
ix
Σ
o
f
th
e
f
ea
tu
r
es
ac
r
o
s
s
a
ll
s
am
p
les
in
a
d
atab
ase.
T
h
e
Ma
h
alan
o
b
is
d
is
tan
ce
is
g
iv
en
as (
3
)
[
31
]
.
(
,
)
=
(
−
)
∑
(
−
)
−
1
(
3
)
T
h
e
m
ah
alan
o
b
is
d
is
tan
ce
m
etr
ic
is
s
ca
led
ac
co
r
d
in
g
to
th
e
p
r
ec
is
io
n
m
atr
ix
(
th
e
co
v
a
r
ian
c
e
m
atr
ix
's
in
v
er
s
e)
[
32
]
.
I
t
p
r
o
v
i
d
es
a
way
o
f
r
ed
u
ci
n
g
t
h
e
in
f
l
u
en
c
e
o
f
d
is
tan
ce
s
alo
n
g
d
im
en
s
i
o
n
s
ir
r
elev
a
n
t
to
t
h
e
cu
r
r
en
t
d
escr
ip
tiv
e
wo
r
d
a
n
d
n
o
r
m
alizin
g
d
is
tan
ce
s
ac
r
o
s
s
d
if
f
er
en
t
f
ea
tu
r
e
s
p
ac
es
to
c
r
e
ate
a
s
in
g
le
d
is
tan
ce
v
alu
e
f
o
r
o
b
ject
class
if
icatio
n
.
T
h
e
m
ah
alan
o
b
is
d
is
tan
ce
m
etr
ic
ca
n
b
e
s
ee
n
as
a
f
ea
tu
r
e
weig
h
tin
g
with
in
d
im
en
s
io
n
s
o
f
f
ea
tu
r
es
an
d
e
x
clu
s
iv
e
f
ea
tu
r
es.
Fo
r
ex
am
p
l
e,
th
e
lig
h
tn
ess
d
im
en
s
io
n
o
f
co
lo
r
s
p
ac
e
v
ar
ies
m
o
r
e
t
h
an
th
e
c
o
lo
r
d
im
en
s
io
n
s
f
o
r
a
g
iv
en
co
l
o
r
wo
r
d
.
T
h
er
ef
o
r
e,
d
is
tan
ce
i
n
th
e
lig
h
t
n
ess
d
im
en
s
io
n
h
as
a
r
ed
u
ce
d
ef
f
ec
t
o
n
class
if
icatio
n
.
Scalin
g
f
ea
tu
r
es
in
th
is
m
an
n
er
also
allo
w
u
s
to
co
m
b
in
e
d
is
jo
in
t
f
ea
tu
r
es
o
f
v
ar
y
in
g
d
im
en
s
io
n
s
an
d
d
is
tr
i
b
u
tio
n
s
,
allo
win
g
g
r
ea
ter
f
lex
i
b
ilit
y
f
o
r
f
u
tu
r
e
f
ea
tu
r
es
[
33
]
.
Ma
h
alan
o
b
is
d
is
tan
ce
is
ess
en
tially
a
d
is
tan
ce
m
ea
s
u
r
e
b
ased
o
n
co
r
r
elatio
n
s
b
etwe
en
v
ar
iab
les
b
y
wh
ich
d
if
f
er
e
n
t
p
atter
n
s
ca
n
b
e
id
en
tifie
d
an
d
an
aly
ze
d
.
I
t
is
a
u
s
ef
u
l
way
o
f
d
eter
m
i
n
in
g
t
h
e
s
im
ilar
ity
o
f
an
u
n
k
n
o
wn
s
am
p
le
s
et
to
a
k
n
o
wn
o
n
e.
Dis
tan
ce
-
b
ased
ap
p
r
o
ac
h
es
ca
lcu
late
th
e
d
is
tan
c
e
f
r
o
m
a
p
o
in
t
to
a
p
ar
ticu
lar
p
o
in
t
in
th
e
d
ata
s
e
t.
Dis
tan
ce
to
th
e
m
ea
n
,
t
h
e
a
v
er
ag
e
d
is
tan
ce
b
etwe
en
t
h
e
q
u
er
y
p
o
in
t
an
d
all
p
o
in
ts
in
th
e
d
ata
s
et,
th
e
m
ax
im
u
m
d
is
tan
ce
b
etwe
en
th
e
q
u
er
y
p
o
in
t
an
d
d
ata
s
et
p
o
in
ts
ar
e
ex
am
p
les
o
f
th
e
m
an
y
o
p
tio
n
s
.
W
h
eth
er
a
d
ata
p
o
in
t is clo
s
e
to
th
e
d
ata
s
et
d
e
p
en
d
s
o
n
th
e
u
s
er
'
s
th
r
esh
o
ld
[
34
]
.
Ma
h
alan
o
b
is
d
is
tan
ce
is
a
d
is
t
an
ce
b
etwe
en
two
p
o
in
ts
=
(
1
,
2
,
…
,
)
an
d
=
(
1
,
2
,
.
.
,
)
in
th
e
p
d
im
en
s
io
n
al
s
p
ac
e
is
d
ef
in
ed
as (
4
)
[
35
]
.
(
,
)
=
√
(
−
)
−
1
(
−
)
(
4)
W
h
er
e
(
,
0
)
=
‖
‖
=
√
−
1
is
th
e
n
o
r
m
o
f
an
d
−
1
is
a
p
o
s
itiv
e
s
em
i
-
d
ef
in
ite
co
v
ar
ian
ce
m
atr
ic
.
Po
in
ts
with
th
e
s
am
e
d
is
tan
ce
o
f
th
e
o
r
i
g
in
‖
‖
=
s
atis
f
y
−
1
=
2
wh
ich
is
th
e
g
en
er
al
eq
u
atio
n
s
o
f
an
ellip
s
o
id
ce
n
ter
ed
at
th
e
o
r
ig
in
,
an
d
we
ar
e
in
ter
ested
in
th
e
d
is
tan
ce
o
f
a
n
o
b
s
er
v
atio
n
f
r
o
m
its
ce
n
ter
̅
g
iv
en
b
y
(
5
)
.
(
,
̅
)
=
√
(
−
̅
)
−
1
(
−
̅
)
(
5
)
T
h
e
m
ah
alan
o
b
is
d
is
tan
ce
's
d
r
awb
ac
k
is
th
e
eq
u
al
ad
d
in
g
u
p
o
f
th
e
v
a
r
ian
ce
n
o
r
m
alize
d
s
q
u
ar
ed
d
is
tan
ce
s
o
f
th
e
f
ea
tu
r
es.
I
n
t
h
e
ca
s
e
o
f
n
o
is
e
-
f
r
ee
s
ig
n
als,
th
is
lead
s
to
th
e
b
est
p
o
s
s
ib
le
p
er
f
o
r
m
an
ce
.
B
u
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2722
-
2
5
8
6
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
,
Vo
l
.
1
0
,
No
.
4
,
Dec
em
b
er
2
0
2
1
:
3
2
6
–
339
330
s
u
p
p
o
s
e
th
e
f
ea
tu
r
e
is
d
is
to
r
te
d
b
y
n
o
is
e
d
u
e
to
th
e
s
q
u
ar
i
n
g
o
f
th
e
d
is
tan
ce
s
.
I
n
th
at
ca
s
e,
a
s
in
g
le
f
ea
tu
r
e
ca
n
h
av
e
s
u
ch
a
h
ig
h
v
alu
e
th
at
it
co
v
er
s
th
e
o
t
h
er
f
ea
tu
r
es'
in
f
o
r
m
atio
n
an
d
lead
s
to
a
m
is
class
if
icatio
n
[
35
]
.
T
h
er
ef
o
r
e,
to
f
in
d
class
if
icatio
n
p
r
o
ce
d
u
r
es
m
o
r
e
r
o
b
u
s
t
to
n
o
is
e,
we
h
av
e
to
f
in
d
a
d
is
tan
c
e
m
ea
s
u
r
e
th
at
g
iv
es
less
weig
h
t
to
th
e
n
o
is
y
f
ea
tu
r
es
a
n
d
m
o
r
e
weig
h
t
to
th
e
clea
n
f
ea
tu
r
es.
I
t
is
r
ea
ch
ed
b
y
c
o
m
p
ar
i
n
g
th
e
d
if
f
er
en
t
in
p
u
t
f
ea
tu
r
es
to
d
ec
id
e
wh
ich
f
ea
tu
r
e
s
h
o
u
ld
b
e
g
iv
en
less
weig
h
t
o
r
ex
clu
d
ed
an
d
h
av
e
m
o
r
e
weig
h
t
[
36
]
,
[
37
]
.
2
.
2
.
3
.
So
un
dex
di
s
t
a
nce
Ph
o
n
etic
en
co
d
in
g
tech
n
iq
u
es
co
n
s
id
er
a
wo
r
d
p
h
o
n
etic
t
r
a
n
s
cr
ip
tio
n
f
o
r
cl
ass
if
icatio
n
a
n
d
co
d
in
g
p
u
r
p
o
s
es,
s
u
ch
as
co
r
r
ec
tin
g
ev
en
tu
al
s
p
ellin
g
m
is
tak
e
s
an
d
class
if
y
in
g
p
h
o
n
etica
l
ly
d
ig
ital
lib
r
ar
ies,
d
ictio
n
ar
ies,
an
d
d
atab
ases
[
38
]
.
T
h
e
p
h
o
n
etic
r
e
p
r
esen
ta
tio
n
h
as
s
ev
er
al
ap
p
licatio
n
s
.
First,
it
allo
ws
to
s
ea
r
ch
co
n
ce
p
ts
b
ased
o
n
p
r
o
n
u
n
ciatio
n
r
ath
e
r
th
an
s
p
ellin
g
[
38
]
-
[
40
]
.
T
h
e
s
o
u
n
d
ex
p
h
o
n
etic
tech
n
i
q
u
e
was
m
ain
ly
u
s
ed
in
ap
p
licatio
n
s
in
v
o
lv
in
g
s
ea
r
ch
in
g
p
eo
p
le'
s
n
am
es
lik
e
air
r
eser
v
atio
n
s
y
s
tem
s
,
ce
n
s
u
s
es,
an
d
o
th
e
r
task
s
p
r
esen
tin
g
ty
p
in
g
er
r
o
r
s
d
u
e
to
p
h
o
n
etic
s
im
ilar
ity
[
41
]
.
Sch
ü
tze
et
a
l
.
[
4
2
]
r
ep
o
r
ted
th
at
th
e
s
o
u
n
d
ex
tech
n
iq
u
e
ev
alu
ates
ea
ch
letter
in
th
e
in
p
u
t
wo
r
d
an
d
ass
ig
n
s
a
n
u
m
er
ic
v
alu
e
th
at
co
n
v
er
ts
ea
ch
wo
r
d
in
t
o
a
c
o
d
e
m
ad
e
u
p
o
f
f
o
u
r
el
em
en
ts
[
43
]
.
T
h
u
s
,
s
o
u
n
d
ex
u
s
es n
u
m
er
ic
co
d
es f
o
r
ea
ch
letter
o
f
th
e
s
tr
in
g
to
b
e
co
d
if
ied
,
as sh
o
w
n
in
T
a
b
le
1
.
T
ab
le
1
.
So
u
n
d
ex
p
h
o
n
etic
co
d
es f
o
r
E
n
g
lis
h
alp
h
ab
e
t
Nu
m
e
ric Co
d
e
Letter
0
a,
i,
u,
e,
o,
y
1
b
,
p
,
f,
v
2
c
,
g
,
j
,
k
,
q
,
s,
x
,
z
3
d
,
t
4
L
5
m
,
n
6
r
2
.
2
.
4
.
N
-
g
ra
m
s
dis
t
a
nce
An
N
-
g
r
am
is
a
s
u
b
-
s
eq
u
en
ce
o
f
N
item
s
f
r
o
m
a
g
iv
en
s
eq
u
en
ce
.
N
-
g
r
am
s
ar
e
u
s
ed
in
v
ar
io
u
s
ar
ea
s
o
f
s
tatis
tical
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
a
n
d
g
en
etic
s
e
q
u
en
ce
a
n
aly
s
is
.
T
h
e
item
s
in
q
u
esti
o
n
ca
n
b
e
ch
ar
ac
ter
s
,
wo
r
d
s
,
o
r
b
ase
p
air
s
ac
co
r
d
in
g
to
th
e
a
p
p
lica
tio
n
.
Fo
r
ex
am
p
le,
th
is
N
-
g
r
am
o
u
t
p
u
t c
an
b
e
u
s
ed
f
o
r
s
tatis
t
ical
m
ac
h
in
e
tr
an
s
latio
n
an
d
s
p
ell
ch
ec
k
in
g
[
44
]
.
Patter
n
ex
tr
ac
tio
n
is
th
e
p
r
o
ce
s
s
o
f
p
ar
s
in
g
a
s
eq
u
en
ce
o
f
item
s
to
f
in
d
o
r
ex
tr
ac
t
a
p
ar
ticu
lar
p
atter
n
o
f
item
s
.
Patter
n
len
g
th
ca
n
b
e
f
ix
ed
,
as
in
th
e
n
-
g
r
am
m
o
d
el,
o
r
it
ca
n
b
e
v
ar
ia
b
le.
Var
ia
b
le
-
len
g
th
p
atter
n
s
ca
n
b
e
d
ir
ec
tiv
es
to
s
p
ec
if
ic
r
u
les,
lik
e
r
eg
u
lar
ex
p
r
ess
io
n
s
.
Ho
wev
er
,
th
e
y
ca
n
also
b
e
r
a
n
d
o
m
an
d
d
e
p
en
d
o
n
th
e
co
n
tex
t a
n
d
p
atter
n
r
ep
etiti
o
n
in
th
e
p
atter
n
s
d
ictio
n
a
r
y
[
45
]
-
[
47
]
.
3.
P
RO
P
O
SE
D
ADAP
T
I
V
E
L
ANG
UAG
E
P
RO
C
E
SS
I
NG
UNIT
T
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
,
s
h
o
w
n
in
Fig
u
r
e
4
,
in
v
o
lv
es tex
t c
lass
if
ier
s
,
wh
er
e
th
ey
class
if
y
tex
t in
p
u
t to
its
co
r
r
esp
o
n
d
in
g
w
o
r
d
ta
g
g
in
g
.
T
h
e
s
y
s
tem
im
p
lem
en
t
s
NL
P
,
HM
M
,
an
d
B
ay
esi
an
as
an
ad
ap
tiv
e
co
m
b
in
atio
n
m
o
d
u
le.
T
h
e
H
MM
an
d
B
ay
esian
Netwo
r
k
ar
e
im
p
lem
e
n
ted
t
o
g
eth
er
to
co
v
e
r
th
e
v
a
r
io
u
s
len
g
th
s
o
f
th
e
in
p
u
t
tex
t,
in
wh
ich
th
e
B
ay
esian
Netwo
r
k
h
an
d
les
lo
n
g
e
r
s
en
ten
ce
s
,
wh
ile
HM
M
h
an
d
les
s
h
o
r
ter
s
en
ten
ce
s
.
Su
ch
an
ad
a
p
tiv
e
s
elec
tio
n
o
f
class
if
ier
s
in
th
e
p
r
o
p
o
s
ed
s
y
s
tem
allo
ws f
o
r
lo
n
g
er
s
en
ten
ce
s
to
b
e
p
er
f
o
r
m
e
d
ac
cu
r
ately
.
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
lan
g
u
a
g
e
p
r
o
ce
s
s
in
g
u
n
it
co
n
tai
n
s
wo
r
d
id
en
tific
atio
n
an
d
tag
g
i
n
g
an
d
an
ad
ap
tiv
e
class
if
ier
,
au
to
m
atic
ally
s
elec
tin
g
th
e
class
if
ier
,
e
ith
er
HM
M
o
r
B
ay
esian
Netwo
r
k
,
b
ased
o
n
th
e
n
u
m
b
er
o
f
wo
r
d
s
d
etec
ted
.
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
is
e
v
alu
at
ed
b
ased
o
n
its
s
u
cc
ess
r
ate
a
n
d
p
r
o
ce
s
s
in
g
tim
e
.
T
h
e
s
u
cc
ess
r
ate
in
d
icate
s
wh
eth
er
th
e
o
u
tp
u
t
f
r
o
m
th
e
l
an
g
u
ag
e
p
r
o
c
ess
in
g
u
n
it
(
L
P
U)
co
n
tain
s
all
th
e
im
p
o
r
tan
t
wo
r
d
s
with
th
e
co
r
r
ec
t
s
tr
u
ctu
r
e
(
s
u
b
ject
-
p
r
e
d
ica
te
-
o
b
ject
)
o
r
n
o
t.
I
f
all
-
im
p
o
r
t
an
t
wo
r
d
s
d
etec
ted
f
r
o
m
th
e
s
p
ee
ch
ar
e
in
cl
u
d
e
d
in
th
e
o
u
tp
u
t,
th
en
th
e
s
y
s
tem
is
co
n
s
id
er
ed
a
s
u
cc
es
s
.
Oth
er
wis
e,
it
i
s
c
o
n
s
id
er
ed
a
f
ailu
r
e.
T
h
e
p
r
o
c
ess
in
g
tim
e
is
d
ef
in
ed
as
th
e
L
PU
'
s
m
ea
s
u
r
ed
tim
e,
s
tar
tin
g
f
r
o
m
in
p
u
ttin
g
d
ata
u
n
til o
b
tain
in
g
th
e
o
u
tp
u
t f
r
o
m
th
e
L
PU
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t
J
R
o
b
&
A
u
to
m
I
SS
N:
2722
-
2
5
8
6
A
d
a
p
tive
la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
u
n
it fo
r
Ma
la
ysia
n
s
ig
n
la
n
g
u
a
g
e
s
yn
th
esiz
er
(
Ha
r
is
A
l Q
o
d
r
i Ma
a
r
if
)
331
Fig
u
r
e
4
.
B
lo
ck
d
iag
r
am
o
f
th
e
p
r
o
p
o
s
ed
ad
ap
ti
v
e
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
u
n
it
3
.
1
.
N
-
g
ra
m
s
I
n
co
m
p
u
tatio
n
al
lin
g
u
is
tics
,
a
s
eq
u
en
ce
o
f
co
n
tig
u
o
u
s
wo
r
d
s
is
ca
lled
N
-
g
r
a
m
[
46
-
48
]
.
T
h
e
s
y
s
tem
im
p
lem
en
ts
a
c
o
m
b
in
a
tio
n
o
f
n
-
g
r
am
s
an
d
NL
P
.
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
,
L
PU
,
is
b
as
ed
o
n
n
-
g
r
am
s
an
d
NL
P.
Fig
u
r
e
5
s
h
o
ws
th
at
th
e
o
u
tp
u
t
f
r
o
m
N
-
g
r
am
s
is
f
ed
to
th
e
L
PU.
L
et
th
e
in
p
u
t
to
t
h
e
L
PU
b
e
o
n
th
e
g
r
o
u
p
o
f
w
o
r
d
s
(
=
2
)
.
NL
P
s
u
b
s
eq
u
en
tly
p
r
o
ce
s
s
es
th
e
p
r
o
ce
s
s
ed
tex
t
in
p
u
t
to
g
et
th
e
p
r
o
p
er
s
en
ten
ce
s
r
ep
r
esen
ted
b
y
a
p
a
r
ticu
lar
s
ig
n
lan
g
u
ag
e'
s
s
ig
n
s
.
I
n
a
n
N
-
g
r
am
,
s
en
ten
ce
s
ar
e
tr
u
n
ca
te
d
to
th
e
len
g
th
(
−
1
)
an
d
its
tr
u
n
ca
tio
n
p
r
o
b
ab
ilit
y
i
s
d
ef
in
ed
as (
6
)
.
(
|
1
…
,
−
1
)
=
(
|
−
+
1
…
,
−
1
)
(
6
)
Sin
ce
=
2
,
it
h
as
b
ee
n
ca
lled
a
b
ig
r
am
.
I
n
s
u
ch
ca
s
e,
th
e
-
g
r
am
co
n
d
itio
n
al
p
r
o
b
a
b
ilit
ies
(
|
−
1
)
ca
n
b
e
esti
m
ated
f
r
o
m
r
aw
t
ex
t
(
−
1
)
b
ased
o
n
th
e
r
elativ
e
f
r
e
q
u
en
cy
o
f
wo
r
d
s
eq
u
e
n
ce
s
(
−
1
)
as
(
7
)
an
d
(
8
)
.
(
|
−
1
)
=
(
−
1
)
(
−
1
)
(
7
)
(
|
−
+
1
−
1
)
=
(
−
+
1
−
1
)
(
−
+
1
−
1
)
(
8
)
3
.
2
.
Na
t
ura
l
la
ng
ua
g
e
pro
ce
s
s
ing
T
h
e
L
PU
p
er
f
o
r
m
s
th
e
tex
t
-
tr
an
s
lated
p
r
o
ce
s
s
ac
co
r
d
in
g
to
Sig
n
L
an
g
u
a
g
e'
s
g
r
am
m
atica
l
r
u
les,
in
th
is
ca
s
e,
th
e
Ma
lay
s
ian
s
ig
n
lan
g
u
ag
e
(
MSL
)
.
First,
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
i
d
en
tifie
s
th
e
"im
p
o
r
tan
t"
wo
r
d
s
th
at
th
e
p
r
o
p
o
s
ed
SL
s
y
n
th
esizer
s
y
n
th
esizes.
T
h
en
,
it
u
tili
ze
s
th
e
"tag
g
in
g
"
p
r
o
ce
s
s
,
wh
i
ch
lab
els
th
e
in
p
u
t
wo
r
d
in
to
s
p
ec
if
ic
s
tr
u
ct
u
r
e
ca
teg
o
r
ies,
i.e
.
,
s
u
b
ject,
p
r
ed
icat
e,
an
d
o
b
ject
(S
-
P
-
O)
.
NL
P
is
th
e
ess
en
tia
l
p
r
o
ce
s
s
in
t
h
e
L
PU.
Fig
u
r
e
5
s
h
o
w
s
th
e
d
etail
o
f
th
e
NL
P.
I
t
in
v
o
lv
es
a
to
k
en
izer
,
POS
tag
g
er
,
n
am
e
d
en
tity
ex
tr
ac
tio
n
,
s
tem
m
er
,
an
d
lex
ical
tr
an
s
f
er
.
T
h
e
s
tep
s
ar
e
r
eq
u
ir
e
d
to
en
h
an
ce
th
e
tr
an
s
latio
n
an
d
th
e
o
u
tp
u
t
o
f
th
e
s
y
s
tem
.
Fig
u
r
e
5
.
W
o
r
d
p
r
o
ce
s
s
in
g
o
f
t
h
e
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
To
S
en
ten
ce
L
en
g
th
C
a
lcu
la
to
r
Token
i
ze
r
P
O
S
Tagge
r
N
am
ed
Ent
i
t
y
E
xtr
acti
on
St
em
m
er
L
exi
ca
l
Trans
f
er
D
at
aba
s
e
T
ext
I
n
p
u
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2722
-
2
5
8
6
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
,
Vo
l
.
1
0
,
No
.
4
,
Dec
em
b
er
2
0
2
1
:
3
2
6
–
339
332
3
.
2
.
1
.
T
o
k
enizer
a
nd
P
O
S
t
a
g
g
er
T
o
k
en
is
in
g
is
a
b
asic
o
p
e
r
atio
n
o
f
NL
P
th
at
is
a
p
p
lied
to
an
in
p
u
t
tex
t.
I
t
b
r
ea
k
s
u
p
a
s
tr
ea
m
o
f
ch
ar
ac
ter
s
in
to
wo
r
d
s
,
p
u
n
ctu
a
tio
n
m
ar
k
s
,
n
u
m
b
e
r
s
,
an
d
o
th
e
r
d
is
cr
ete
item
s
.
F
ig
u
r
e
6
s
h
o
ws
th
e
f
lo
wch
ar
t
o
f
th
e
to
k
e
n
izer
.
T
h
is
NL
P
s
tag
e
d
o
es
n
o
t
class
if
y
g
r
am
m
ati
ca
l
ca
teg
o
r
ies
o
f
th
e
in
p
u
t
te
x
t.
I
t
also
d
o
es
n
o
t
co
n
s
id
er
a
n
y
in
f
o
r
m
atio
n
o
n
t
h
e
s
y
n
tactic
s
tr
u
ct
u
r
e
o
f
th
e
t
ex
t
o
r
th
e
ty
p
e
o
f
wo
r
d
s
in
it
(
e.
g
.
,
wh
et
h
er
t
h
e
wo
r
d
s
ar
e
v
e
r
b
s
an
d
n
o
u
n
s
)
.
T
h
e
in
p
u
t o
f
th
e
to
k
e
n
izer
is
th
e
id
en
tifie
d
wo
r
d
s
(
tex
t
in
p
u
t)
.
I
n
co
m
p
ar
is
o
n
,
t
h
e
o
u
tp
u
t
is
th
e
co
r
r
esp
o
n
d
in
g
to
k
en
ized
wo
r
d
s
.
T
h
e
last
wo
r
d
is
al
s
o
tag
g
ed
to
in
d
icate
th
at
it
is
th
e
last
to
k
en
f
o
r
th
e
c
u
r
r
e
n
t in
p
u
t,
an
d
u
s
u
a
lly
,
th
e
last
wo
r
d
is
a
n
o
u
n
.
Fig
u
r
e
6
.
Flo
wch
ar
t
o
f
to
k
en
iz
er
T
h
e
POS
tag
g
er
allo
ws
class
i
f
y
in
g
t
h
e
wo
r
d
s
in
to
n
in
e
tr
ad
itio
n
al
wo
r
d
class
es,
i.e
.
,
n
o
u
n
,
v
er
b
,
ad
jectiv
e,
ad
v
er
b
,
p
r
ep
o
s
itio
n
,
ar
ticle,
in
ter
jectio
n
,
p
r
o
n
o
u
n
,
an
d
co
n
ju
n
ctio
n
.
I
n
a
d
d
itio
n
,
ea
ch
wo
r
d
e
n
ter
in
g
th
e
POS
tag
g
in
g
is
lab
eled
ac
co
r
d
in
g
to
its
s
tatu
s
in
a
s
en
te
n
ce
.
T
h
u
s
,
i
t
co
n
tr
ib
u
tes
to
th
e
tr
an
s
latio
n
p
r
o
ce
s
s
d
o
wn
s
tr
ea
m
[
49
]
,
[
50
]
.
E
ac
h
wo
r
d
e
n
ter
in
g
t
h
e
POS
T
ag
g
in
g
b
lo
ck
is
lab
eled
ac
c
o
r
d
in
g
to
its
s
tatu
s
in
a
s
en
ten
c
e.
Fig
u
r
e
7
s
h
o
ws
th
e
ta
g
g
in
g
p
r
o
ce
s
s
wh
er
e
ea
ch
wo
r
d
is
tag
g
ed
ac
c
o
r
d
in
g
to
its
p
ar
ticu
lar
lab
el.
E
a
ch
to
k
en
is
tag
g
ed
with
its
co
r
r
esp
o
n
d
in
g
SP
O
c
ateg
o
r
y
(
i.e
.
,
as
a
s
u
b
ject,
p
r
e
d
icate
,
o
r
o
b
ject
)
.
T
h
e
p
r
ep
o
s
i
tio
n
s
an
d
o
th
er
n
o
n
-
im
p
o
r
tan
t
wo
r
d
s
ar
e
n
o
t
tag
g
e
d
an
d
th
u
s
ar
e
d
is
ca
r
d
ed
.
Fo
r
th
e
s
am
e
ex
am
p
le
as
b
ef
o
r
e
,
th
e
tag
g
ed
to
k
en
as
to
k
en
1
as
"S"
(
s
u
b
ject
)
,
to
k
en
2
as
"P"
(
p
r
ed
icate
)
,
to
k
en
3
as
"O"
(
o
b
ject
)
,
to
k
en
4
as
"
u
n
k
n
o
wn
,
"
an
d
to
k
en
5
as "O
"
(
o
b
ject
).
Fig
u
r
e
7
.
Flo
wch
ar
t
o
f
POS
tag
g
er
3
.
2
.
2
.
Na
m
ed
ent
it
y
ex
t
ra
ct
io
n
Nam
ed
en
tity
ex
tr
ac
tio
n
id
en
tifie
s
ty
p
es
o
f
tag
g
ed
wo
r
d
s
,
s
u
ch
as
n
am
es
o
f
p
er
s
o
n
s
,
v
er
b
s
,
an
d
r
em
o
v
es
wo
r
d
s
with
u
n
k
n
o
wn
tag
g
ed
t
o
k
en
s
,
as
s
h
o
w
n
in
Fig
u
r
e
8
.
T
h
e
s
y
n
tactic
an
aly
s
is
allo
ws
th
e
id
en
tific
atio
n
o
f
th
e
SP
O
s
y
n
t
ac
tic
co
m
p
o
n
en
ts
o
f
t
h
e
s
en
te
n
ce
.
Fo
r
e
x
am
p
le,
if
th
e
t
o
k
en
ized
wo
r
d
is
a
n
o
u
n
,
th
e
s
y
n
tactic
an
aly
s
is
v
alid
ates th
e
ap
p
r
o
p
r
iaten
ess
o
f
th
e
ta
g
g
iv
en
to
th
e
to
k
en
ized
w
o
r
d
s
.
Fig
u
r
e
8
.
Flo
wch
ar
t
o
f
th
e
n
a
m
ed
en
tity
ex
tr
a
ctio
n
an
d
s
y
n
t
ac
tic
an
aly
ze
r
S
tart
T
ext
Inp
u
t
T
ok
enizing
Ea
ch
W
o
rd
L
abelli
ng
T
o
k
en
End
S
ta
rt
T
agg
ing
T
o
k
en
Inp
u
t
End
Database
L
a
st
T
oke
n?
Receive
Inp
u
t
T
o
k
en
Y
es
No
Start
T
ag
g
ed
T
o
k
en
Inp
u
t
Sy
n
tactic
An
aly
zer
End
V
e
r
ify
T
a
g
g
e
d
T
oke
n?
Rem
o
v
al
T
o
k
en
= “
Un
k
n
o
wn
”
Y
es
No
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t
J
R
o
b
&
A
u
to
m
I
SS
N:
2722
-
2
5
8
6
A
d
a
p
tive
la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
u
n
it fo
r
Ma
la
ysia
n
s
ig
n
la
n
g
u
a
g
e
s
yn
th
esiz
er
(
Ha
r
is
A
l Q
o
d
r
i Ma
a
r
if
)
333
Fo
r
th
e
s
am
e
ex
am
p
le,
th
e
p
r
o
ce
s
s
ed
to
k
en
b
y
NE
E
(
af
ter
POS
tag
g
er
)
is
r
esu
lted
as
to
k
en
1
as
"S"
(
s
u
b
ject
)
,
to
k
e
n
2
as
"P"
(
p
r
ed
icate
)
,
to
k
en
3
as
"O"
(
o
b
ject
)
,
an
d
to
k
en
5
as
"O"
(
o
b
ject
)
.
T
h
en
,
af
ter
p
ass
in
g
th
e
Sy
n
tactic
An
al
y
ze
r
p
r
o
ce
s
s
,
th
e
r
esu
lted
to
k
e
n
is
to
k
en
1
as
"O"
(
o
b
ject
)
,
to
k
e
n
2
as
"P
"
(
p
r
e
d
icate
)
,
t
o
k
en
3
as "S"
(
s
u
b
ject
)
,
an
d
to
k
en
5
as "S"
(
s
u
b
ject
).
3
.
2
.
3
.
Ste
m
m
er
a
nd
lex
ica
l t
ra
ns
f
er
T
h
e
s
tem
m
er
is
u
s
ed
to
id
e
n
tify
b
asic
f
o
r
m
s
(
s
tems
)
o
f
wo
r
d
s
allo
win
g
to
in
f
er
g
en
d
er
in
f
o
r
m
atio
n
an
d
th
e
n
u
m
b
e
r
o
f
i
n
p
u
t
wo
r
d
s
.
I
t
is
aim
ed
to
m
ap
a
s
p
ee
c
h
(
tex
t)
in
a
p
ar
ticu
lar
lan
g
u
ag
e
to
a
co
r
r
esp
o
n
d
in
g
s
ig
n
in
th
e
tar
g
et
SL.
I
t
u
s
es
a
lan
g
u
ag
e
d
ictio
n
ar
y
to
p
e
r
f
o
r
m
an
ac
cu
r
ate
r
ed
u
ctio
n
to
r
o
o
t
wo
r
d
s
.
Fig
u
r
e
9
s
h
o
ws
th
e
f
lo
wch
ar
t
f
o
r
th
e
s
tem
m
in
g
p
r
o
ce
s
s
.
Stem
m
in
g
u
s
es
o
r
d
in
ar
y
p
atter
n
m
atch
in
g
to
s
tr
ip
s
u
f
f
ix
es
o
f
to
k
en
s
s
im
p
ly
(
e.
g
.
,
r
e
m
o
v
e
"
-
s
,
"
an
d
r
em
o
v
e
"
-
i
n
g
,
",
in
th
e
wo
r
d
en
d
in
g
s
)
,
th
u
s
"stri
p
p
in
g
o
f
f
"
ty
p
ical
g
r
am
m
ar
.
I
n
th
e
p
r
o
p
o
s
ed
s
y
s
tem
,
th
e
s
tem
m
er
is
u
s
ed
to
id
en
tify
v
e
r
b
s
o
n
ly
,
tag
g
e
d
as
P
r
e
d
icate
.
T
h
is
s
tag
e
is
les
s
u
s
ed
in
s
o
m
e
lan
g
u
ag
e
s
(
in
clu
d
in
g
B
ah
asa
Me
lay
u
)
,
wh
er
e
n
o
ten
s
e
-
d
ep
e
n
d
en
t
ch
a
n
g
es
o
f
r
o
o
t
wo
r
d
s
ar
e
n
ee
d
ed
.
Ho
wev
er
,
af
f
ix
es
m
ay
b
e
u
s
ed
to
g
iv
e
ex
tr
a
e
m
p
h
asis
to
th
e
m
ea
n
in
g
o
f
th
e
r
o
o
t
wo
r
d
s
.
Als
o
,
th
ey
m
ig
h
t
b
e
a
p
p
lied
to
d
e
r
iv
e
n
ew
wo
r
d
s
(
u
s
u
ally
-
v
e
r
b
s
)
t
h
at
h
av
e
d
if
f
e
r
en
t
m
ea
n
i
n
g
s
th
o
u
g
h
s
till
r
elate
to
th
e
r
o
o
t
o
n
es.
Fo
r
ex
am
p
le,
t
o
k
en
2
"d
im
a
k
an
"
is
tag
g
ed
as
"P."
T
h
e
tag
g
in
g
"P"
r
ef
e
r
s
to
th
e
v
er
b
,
wh
e
r
e
it
r
em
o
v
es
"d
i"
as
a
p
r
ef
ix
f
o
r
t
h
e
wo
r
d
"m
ak
a
n
.
"
to
k
en
2
c
h
an
g
es
in
to
"m
ak
an
"
an
d
with
th
e
s
am
e
tag
"P"
b
y
h
av
in
g
t
h
is
p
r
o
ce
s
s
.
T
h
e
lex
ical
tr
an
s
f
er
in
v
o
lv
es
o
n
e
-
to
-
o
n
e
m
ap
p
i
n
g
o
f
th
e
in
p
u
t
s
en
ten
ce
s
to
t
h
eir
co
r
r
esp
o
n
d
in
g
m
ea
n
in
g
.
I
t
r
eq
u
ir
es
r
ef
er
r
in
g
to
t
h
e
d
ictio
n
ar
y
an
d
wo
r
d
d
atab
ase,
s
ee
in
Fig
u
r
e
1
0
.
Sp
ec
if
ically
,
th
e
s
tag
e
allo
ws
d
is
tin
g
u
is
h
in
g
wo
r
d
s
h
av
in
g
m
u
ltip
le
m
ea
n
in
g
s
.
I
f
th
e
wo
r
d
s
h
av
e
two
o
r
m
o
r
e
m
ea
n
i
n
g
s
,
th
e
m
ea
n
in
g
b
ased
o
n
t
h
e
SP
O
tag
g
in
g
in
f
o
r
m
atio
n
o
f
t
h
e
s
en
ten
ce
is
s
elec
ted
.
Fig
u
r
e
9
.
Flo
wch
ar
t
o
f
th
e
s
tem
m
in
g
p
r
o
ce
s
s
Fig
u
r
e
1
0
.
Flo
wch
ar
t o
f
th
e
le
x
ical
tr
an
s
f
er
3
.
3
.
Sente
nce
leng
t
h c
a
lcula
t
o
r
a
nd
a
da
ptiv
e
s
elec
t
io
n o
f
t
he
cla
s
s
if
ier
T
h
e
s
en
ten
ce
le
n
g
th
ca
lcu
lato
r
(
SLC)
is
a
s
tr
in
g
p
r
o
ce
s
s
in
g
th
at
ca
lcu
lates
th
e
n
u
m
b
e
r
o
f
wo
r
d
s
in
a
s
en
ten
ce
.
I
n
th
is
r
esear
ch
,
an
y
in
p
u
t g
iv
e
n
to
th
e
s
y
s
tem
s
is
lim
ited
to
b
e
o
n
e
co
m
p
lete
s
en
ten
ce
,
an
d
t
h
e
wo
r
d
co
u
n
t
is
th
e
n
u
m
b
er
o
f
wo
r
d
s
in
th
e
in
p
u
t
s
en
ten
ce
.
Fig
u
r
e
1
1
s
h
o
ws
th
e
f
lo
wch
ar
t
f
o
r
th
e
SLC.
I
n
ad
d
itio
n
,
th
e
p
r
o
p
o
s
ed
s
y
s
tem
o
f
f
er
s
an
ad
ap
tiv
e
s
elec
tio
n
o
f
th
e
clas
s
if
ier
,
allo
win
g
f
o
r
s
witch
in
g
b
etwe
en
HM
M
an
d
B
ay
esian
n
etwo
r
k
s
au
to
m
atica
lly
.
T
h
e
HM
M
an
d
B
ay
esian
Netwo
r
k
id
en
tif
y
th
e
s
en
ten
ce
an
d
p
r
o
ce
s
s
it
s
u
ch
th
at
it
f
o
llo
ws
th
e
p
r
e
-
ass
ig
n
ed
o
r
d
er
,
i.e
.
,
s
u
b
ject,
p
r
ed
icate
,
an
d
o
b
ject
(
SP
O)
,
b
ased
o
n
th
e
n
u
m
b
er
o
f
wo
r
d
s
in
th
e
s
en
ten
ce
.
B
ased
o
n
th
e
ex
p
e
r
im
en
tal
r
esu
lt,
HM
M
wo
r
k
s
f
o
r
s
h
o
r
t
s
en
ten
ce
s
(
th
r
esh
o
ld
=
7
w
o
r
d
s
)
,
a
n
d
B
ay
esian
n
etwo
r
k
s
ap
p
lies
f
o
r
lo
n
g
e
r
s
en
ten
ce
s
(
m
o
r
e
th
an
7
wo
r
d
s
)
.
On
ce
th
e
r
esu
ltin
g
s
en
ten
ce
h
as
b
ee
n
ar
r
an
g
e
d
in
to
th
e
SP
O
o
r
d
er
,
th
e
s
en
ten
ce
is
f
ed
in
to
th
e
a
n
im
atio
n
p
a
r
t,
wh
ich
allo
ws th
e
an
im
ated
av
atar
.
S
ta
rt
T
oke
n
a
s
V
e
r
b?
W
ord =
(
Pre
f
ix
)
-
St
e
m
-
(
Suf
f
ix
)
C
he
c
k
P
re
fix
R
e
move
P
re
fix
C
he
c
k
Suf
f
ix
Re
move
Suf
f
ix
End
R
e
c
e
ive
I
nput
T
oke
n
Y
es
No
Y
es
No
No
Y
es
No
S
ta
rt
Receive
Inp
u
t
T
o
k
en
Ch
eck M
eanin
g
o
f
The
W
o
rds
D
a
ta
ba
se
Select
Pr
o
p
er
Meanin
g
En
d
L
a
st
T
oke
n?
Y
es
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2722
-
2
5
8
6
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
,
Vo
l
.
1
0
,
No
.
4
,
Dec
em
b
er
2
0
2
1
:
3
2
6
–
339
334
Fig
u
r
e
1
1
.
Flo
wch
ar
t o
f
th
e
s
en
ten
ce
len
g
th
ca
lcu
la
to
r
HM
M
s
tate
s
an
d
tr
an
s
itio
n
m
atr
ix
ar
e
d
esig
n
ed
to
o
r
g
an
ize
r
esu
lts
in
to
Su
b
ject,
Pre
d
icate
,
an
d
Ob
ject
p
atter
n
s
.
HM
M
in
v
o
lv
es
th
r
ee
s
tates
an
d
a
s
in
g
le
o
u
tp
u
t.
I
t
h
as
t
h
r
ee
o
b
s
er
v
atio
n
p
r
o
b
ab
ilit
y
d
is
tr
ib
u
tio
n
s
B
a
n
d
th
r
ee
s
tate
tr
an
s
itio
n
p
r
o
b
ab
ilit
i
es
A
.
f
o
r
ea
ch
s
tate,
em
its
a
s
in
g
l
e
o
u
tp
u
t
.
T
h
e
m
ath
em
atica
l m
o
d
el
is
d
escr
ib
ed
as f
o
llo
ws:
a)
λ
=
{
A
,
B
,
π
}
b)
=
{
11
,
12
,
22
,
23
,
33
}
c)
=
{
,
,
}
T
h
e
in
itial c
o
n
d
itio
n
π
is
d
ef
in
ed
as th
e
b
asic w
o
r
d
s
,
wh
ich
,
i
n
o
u
r
ca
s
e,
h
as th
r
ee
wo
r
d
s
,
i.e
.
,
s
u
b
ject
(
S),
p
r
ed
icate
(
P),
an
d
o
b
ject
(
O)
.
"
s
u
b
ject
"
is
d
ef
in
ed
as
a
p
er
s
o
n
o
r
s
o
m
eth
i
n
g
wh
ic
h
d
o
es
th
e
ac
tio
n
.
"Pr
ed
icate
"
is
class
if
ied
as
an
ac
tiv
e
v
er
b
th
at
i
n
d
icate
s
th
e
ac
tiv
ity
wh
ich
is
d
o
n
e
b
y
t
h
e
s
u
b
ject.
Fin
ally
,
"o
b
ject"
is
class
if
ied
as
a
p
er
s
o
n
o
r
o
b
ject,
w
h
ich
is
t
h
e
s
u
b
j
ec
t's
g
o
al.
T
h
e
SP
O
s
tr
u
ctu
r
e
i
s
p
r
o
p
o
s
ed
to
ea
s
e
th
e
s
y
n
th
esizer
p
r
o
ce
s
s
,
wh
er
e
th
e
co
m
p
lete
s
en
ten
ce
s
h
o
u
ld
co
n
s
is
t
o
f
th
ese
th
r
ee
b
asic e
le
m
en
ts
SP
O
.
T
h
e
B
ay
esian
n
etwo
r
k
is
ap
p
li
ed
to
a
co
n
d
itio
n
wh
er
e
th
e
n
u
m
b
er
o
f
wo
r
d
s
is
m
o
r
e
th
a
n
7
.
T
h
en
,
th
e
jo
in
t
p
r
o
b
ab
ilit
y
d
is
tr
ib
u
tio
n
r
ep
r
esen
ts
th
e
im
p
lem
e
n
tatio
n
o
f
th
e
B
ay
esian
n
etwo
r
k
.
I
n
th
is
ca
s
e,
th
e
p
r
o
b
a
b
ilis
tic
d
is
tr
ib
u
tio
n
is
t
h
e
r
elatio
n
s
h
ip
am
o
n
g
s
u
b
ject
(
S),
p
r
ed
icate
(
P),
an
d
o
b
ject
(
O)
.
T
h
e
jo
in
t
p
r
o
b
a
b
ilit
y
d
is
tr
ib
u
tio
n
ca
n
b
e
wr
itten
as (
9
)
.
(
,
,
)
=
(
)
(
/
)
(
|
,
)
(
9
)
T
h
e
C
PD is ca
lcu
lated
f
r
o
m
th
e
jo
in
t p
r
o
b
ab
ilit
y
,
an
d
th
e
B
a
y
esian
n
etwo
r
k
c
o
n
s
is
ts
o
f
class
v
ar
iab
les an
d
f
ea
tu
r
e
v
ar
ia
b
les th
at
ar
e
r
ea
d
i
ly
ap
p
licab
le
to
t
h
e
class
if
icati
o
n
task
.
T
h
e
S (
s
u
b
ject
)
is
s
elec
ted
as th
e
class
v
ar
iab
le,
an
d
th
e
ca
lcu
latio
n
f
o
r
th
e
p
r
o
b
a
b
ilit
y
o
f
=
g
iv
en
a
n
y
o
b
s
er
v
e
d
v
alu
e
s
et
(
,
)
as
(
1
0
)
.
(
=
|
,
)
=
(
=
,
,
)
(
=
,
,
)
+
(
=
,
,
)
(
1
0
)
W
h
er
e
(
=
|
,
)
an
d
(
=
,
,
)
ca
n
b
e
c
o
m
p
u
te
d
ef
f
icien
tly
u
s
in
g
(
9
)
.
Similar
ly
,
I
t
ca
n
b
e
a
p
p
lied
to
ca
lcu
late
(
=
,
,
)
.
T
h
en
th
e
v
alu
e
o
f
S
is
d
eter
m
in
ed
b
y
co
m
p
u
tin
g
(
=
,
,
)
an
d
(
=
,
,
)
.
On
ce
th
e
r
esu
ltin
g
s
en
ten
ce
is
ar
r
an
g
ed
in
to
th
e
SP
O
o
r
d
er
,
it
is
f
ed
in
to
th
e
an
im
atio
n
av
atar
(
o
r
r
o
b
o
tic
m
a
n
ip
u
lato
r
)
,
o
u
tp
u
tti
n
g
it in
th
e
c
h
o
s
en
SL
lex
is
.
4.
E
XP
E
R
I
M
E
N
T
A
L
RE
SUL
T
S
AND
D
I
SC
USS
I
O
N
4
.
1
.
E
x
perim
ent
a
l
s
et
up
T
h
e
d
ata
u
s
ed
is
tex
t
-
b
ased
in
p
u
t,
wh
ich
is
co
n
s
id
er
e
d
a
s
im
p
le
Ma
lay
s
en
ten
ce
.
Ho
wev
er
,
co
m
b
in
in
g
th
e
n
u
m
b
er
o
f
wo
r
d
s
in
o
n
e
s
en
ten
ce
,
f
r
o
m
th
r
ee
to
ten
,
in
d
icate
s
in
p
u
t
s
en
t
en
ce
s
f
r
o
m
s
im
p
le
s
en
ten
ce
s
in
to
m
o
r
e
co
m
p
lex
s
en
ten
ce
s
.
T
ab
le
2
s
h
o
ws 1
3
s
am
p
le
d
ata
s
elec
ted
f
r
o
m
th
e
1
3
0
d
ata
u
s
ed
in
th
is
ex
p
er
im
en
t.
T
h
e
s
im
p
le
s
tr
u
ctu
r
e
co
n
s
is
ts
o
f
o
n
ly
SP
O
s
tr
u
ctu
r
e,
wh
er
ea
s
th
e
m
o
r
e
co
m
p
lex
s
en
ten
ce
s
h
av
e
SP
O
s
tr
u
ctu
r
e
an
d
r
a
n
d
o
m
o
r
d
er
o
f
s
en
te
n
ce
s
tr
u
ctu
r
es.
T
h
e
latter
is
in
tr
o
d
u
ce
d
t
o
e
v
alu
ate
th
e
p
r
o
p
o
s
ed
SL
s
y
n
th
esizer
tech
n
iq
u
e
b
ased
o
n
th
e
p
r
o
ce
s
s
in
g
tim
e
an
d
s
u
cc
ess
r
ate.
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
h
as
b
e
en
d
e
v
elo
p
ed
u
s
in
g
Ma
tlab
2
0
1
8
a
an
d
r
u
n
n
i
n
g
at
I
n
tel
i5
5
2
0
0
U
p
r
o
ce
s
s
o
r
an
d
4
GB
R
AM
to
p
r
o
ce
s
s
th
e
o
u
tp
u
t f
r
o
m
s
p
ee
ch
d
ata.
T
o
ev
alu
ate
th
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
s
y
s
tem
,
we
p
er
f
o
r
m
e
d
f
o
u
r
ty
p
es
o
f
co
m
p
ar
is
o
n
s
.
T
h
e
f
ir
s
t
co
m
p
ar
is
o
n
in
v
o
lv
es
ev
alu
atin
g
f
o
u
r
d
if
f
er
en
t
e
d
it
d
is
tan
ce
m
eth
o
d
s
,
n
am
el
y
L
ev
en
s
h
tein
,
s
o
u
n
d
ex
,
N
-
g
r
a
m
s
,
an
d
m
ah
alan
o
b
is
.
T
h
e
s
ec
o
n
d
co
m
p
ar
is
o
n
co
n
s
is
ts
o
f
th
e
s
elec
tio
n
o
f
th
e
n
u
m
b
e
r
o
f
p
ar
a
m
eter
s
u
s
ed
f
o
r
th
e
s
elec
ted
ed
it d
is
tan
ce
.
T
h
e
n
ex
t
co
m
p
ar
is
o
n
is
m
ad
e
to
ev
alu
ate
th
e
NL
P
an
d
HM
M'
s
p
er
f
o
r
m
an
ce
w
ith
NL
P
an
d
B
ay
esian
Netw
o
r
k
.
T
h
e
n
,
we
ca
n
St
a
r
t
Rece
i
v
i
ng
Token
L
a
st
T
a
g
?
L
i
s
t
i
ng
Token,
L
abe
l
s
Toke
n
Count
i
ng
N
um
ber
of R
ece
i
v
ed
T
ok
en
s
En
d
Y
es
No
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t
J
R
o
b
&
A
u
to
m
I
SS
N:
2722
-
2
5
8
6
A
d
a
p
tive
la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
u
n
it fo
r
Ma
la
ysia
n
s
ig
n
la
n
g
u
a
g
e
s
yn
th
esiz
er
(
Ha
r
is
A
l Q
o
d
r
i Ma
a
r
if
)
335
d
eter
m
in
e
th
e
t
h
r
esh
o
ld
n
u
m
b
er
o
f
wo
r
d
s
in
th
e
in
p
u
t
s
e
n
ten
ce
s
f
o
r
th
e
ad
ap
tiv
e
class
if
icatio
n
.
Fin
ally
,
th
e
last
ev
alu
atio
n
is
o
n
th
e
e
d
it d
i
s
tan
ce
s
with
th
e
ad
ap
tiv
e
s
y
s
tem
o
f
th
e
class
if
ier
.
T
ab
le
2
.
Sam
p
le
in
p
u
t
s
elec
ted
f
o
r
la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
u
n
it
No
S
e
l
e
c
t
e
d
I
n
p
u
t
N
u
mb
e
r
o
f
W
o
r
d
s
1
S
a
y
a
M
a
k
a
n
N
a
s
i
(
I
e
a
t
ri
c
e
)
3
2
N
a
si
S
a
y
a
M
a
k
a
n
(
R
i
c
e
I
e
a
t
)
3
3
D
i
a
M
a
k
a
n
N
a
si
(
H
e
/
s
h
e
e
a
t
s r
i
c
e
)
3
4
D
i
a
P
e
r
g
i
k
e
P
a
s
a
r
(
H
e
/
sh
e
g
o
e
s
t
o
M
a
rk
e
t
)
4
5
K
e
P
a
sar
S
a
y
a
P
e
r
g
i
(
T
o
m
a
r
k
e
t
I
g
o
)
4
6
D
i
a
M
a
k
a
n
N
a
si
d
i
P
a
s
a
r
(
H
e
/
s
h
e
e
a
t
s
ri
c
e
a
t
m
a
rk
e
t
)
5
7
S
a
y
a
d
a
n
D
i
a
M
a
k
a
n
N
a
s
i
(
I
a
n
d
H
e
e
a
t
ri
c
e
)
5
8
N
a
si
S
a
y
a
d
a
n
D
i
a
M
a
k
a
n
(
R
i
c
e
I
a
n
d
h
e
/
s
h
e
e
a
t
)
5
9
S
a
y
a
D
i
a
P
e
r
g
i
M
a
k
a
n
P
a
sar
N
a
si
(
I
,
h
e
g
o
e
a
t
m
a
rke
t
r
i
c
e
)
6
10
S
a
y
a
M
a
k
a
n
N
a
s
i
d
a
n
P
e
r
g
i
k
e
P
a
sar
(
I
e
a
t
r
i
c
e
a
n
d
g
o
t
o
m
a
rk
e
t
)
7
11
S
a
y
a
M
a
k
a
n
N
a
s
i
D
a
n
M
e
mb
e
l
i
B
u
a
h
D
i
P
a
s
a
r
(
I
E
a
t
Ri
c
e
a
n
d
B
u
y
a
Fru
i
t
a
t
T
h
e
M
a
rk
e
t
)
8
12
S
a
y
a
d
a
n
D
i
a
M
a
k
a
n
N
a
s
i
d
a
n
B
u
a
h
D
i
P
a
sar
(
I
a
n
d
H
e
/
S
h
e
E
a
t
Ri
c
e
a
n
d
Fru
i
t
i
n
T
h
e
Ma
r
k
e
t
)
9
13
S
a
y
a
M
a
k
a
n
N
a
s
i
d
a
n
P
e
r
g
i
k
e
P
a
sar
d
i
P
a
g
i
H
a
r
i
(
I
E
a
t
R
i
c
e
a
n
d
G
o
t
o
t
h
e
Ma
rke
t
i
n
T
h
e
M
o
r
n
i
n
g
)
10
4
.
2
.
Select
io
n o
f
edit
dis
t
a
nce
a
nd
it
s
cha
ra
ct
er
is
t
ics
E
ac
h
ed
it
d
is
tan
ce
is
u
s
ed
in
t
h
e
p
r
o
p
o
s
ed
s
y
s
tem
,
an
d
th
e
p
er
f
o
r
m
an
ce
o
f
ea
c
h
is
co
m
p
ar
ed
b
ased
o
n
its
s
u
cc
ess
r
ate
an
d
p
r
o
ce
s
s
in
g
tim
e.
T
h
e
s
elec
tio
n
is
p
er
f
o
r
m
ed
to
d
eter
m
in
e
th
e
m
o
s
t
ef
f
ec
tiv
e
ed
it
d
is
tan
ce
an
d
its
ch
ar
ac
ter
is
tics
v
alu
e.
T
h
e
ed
it
d
is
tan
ce
,
w
h
ich
p
r
o
v
id
es
t
h
e
h
i
g
h
est
s
u
cc
ess
r
ate
an
d
th
e
s
h
o
r
test
p
r
o
ce
s
s
in
g
tim
e,
is
s
e
lecte
d
.
T
h
e
e
v
alu
atio
n
is
ca
r
r
i
ed
o
u
t
b
y
co
m
p
a
r
in
g
th
e
p
r
o
c
ess
in
g
tim
e
o
f
ea
c
h
ed
it
d
is
tan
ce
.
I
t
is
n
ec
es
s
ar
y
f
o
r
g
ettin
g
a
f
ast
r
esp
o
n
s
e
f
o
r
th
e
wh
o
le
s
y
s
tem
f
o
r
th
e
SL
s
y
n
th
esizer
.
T
h
e
co
m
p
ar
is
o
n
is
p
er
f
o
r
m
e
d
b
y
v
ar
y
in
g
th
e
n
u
m
b
er
o
f
wo
r
d
s
in
a
s
en
ten
ce
.
T
h
e
co
m
p
ar
is
o
n
aim
s
t
o
f
i
n
d
th
e
s
ig
n
if
ican
t
ed
it
d
is
tan
ce
to
s
u
p
p
o
r
t
t
h
e
SL
s
y
n
th
es
izer
'
s
p
r
o
p
o
s
ed
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
u
n
it.
I
n
th
is
r
esear
ch
,
d
eter
m
in
in
g
th
e
m
o
s
t
s
u
itab
le
ed
it
d
is
tan
ce
tech
n
iq
u
e
is
k
e
y
f
o
r
th
e
p
r
o
p
o
s
ed
s
y
s
tem
to
p
r
o
v
id
e
th
e
f
astes
t
p
r
o
ce
s
s
in
g
tim
e
an
d
th
e
h
ig
h
e
s
t su
cc
ess
r
ate.
Fig
u
r
e
1
2
s
h
o
ws
th
e
p
er
f
o
r
m
an
ce
o
f
ea
ch
ed
it
d
is
tan
ce
with
th
e
v
a
r
io
u
s
n
u
m
b
er
o
f
wo
r
d
s
in
a
s
en
ten
ce
.
I
t
in
d
icate
s
th
at
L
ev
en
s
h
tein
,
s
o
u
n
d
ex
,
an
d
N
-
g
r
am
s
p
r
o
v
id
ed
s
im
ilar
p
r
o
c
ess
in
g
tim
es.
T
h
e
s
im
ilar
ity
o
f
p
r
o
ce
s
s
in
g
tim
e
r
esu
ltin
g
f
r
o
m
th
r
ee
ed
it
d
is
tan
ce
s
i
s
ex
p
ec
ted
d
u
e
to
th
o
s
e
s
im
ilar
ch
ar
ac
ter
is
tics
.
As
n
-
g
r
am
s
h
av
e
p
r
o
v
id
e
d
th
e
f
astes
t
p
r
o
ce
s
s
in
g
tim
e
th
an
m
ah
ala
n
o
b
is
an
d
s
o
u
n
d
ex
d
is
tan
ce
,
N
-
g
r
am
s
h
a
v
e
b
ee
n
s
elec
ted
to
b
e
im
p
lem
e
n
ted
in
t
h
e
p
r
o
p
o
s
ed
s
y
s
tem
'
s
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
u
n
it.
Fig
u
r
e
1
3
s
h
o
ws
th
e
tim
e
c
o
m
p
ar
is
o
n
f
o
r
L
e
v
en
s
h
tein
,
s
o
u
n
d
e
x
,
a
n
d
n
-
g
r
am
s
.
L
ev
en
s
h
tein
an
d
s
o
u
n
d
ex
d
is
tan
ce
r
esu
lted
in
a
lo
n
g
er
p
r
o
ce
s
s
in
g
tim
e
t
h
an
n
-
g
r
a
m
s
d
is
tan
ce
.
I
t
is
d
em
o
n
s
tr
ated
th
at
L
ev
en
s
tein
,
s
o
u
n
d
ex
,
an
d
m
a
h
alan
o
b
is
h
a
v
e
a
lo
n
g
er
p
r
o
ce
s
s
in
g
tim
e
th
an
n
-
g
r
am
s
.
N
-
g
r
am
s
,
at
n
=
2
,
p
r
o
v
id
e
ef
f
icien
t
p
r
o
ce
s
s
in
g
ti
m
e
co
m
p
a
r
ed
to
o
th
er
ed
it
d
is
tan
ce
s
.
T
h
er
ef
o
r
e,
N
-
g
r
am
s
ar
e
im
p
lem
en
ted
f
o
r
th
e
lan
g
u
a
g
e
p
r
o
ce
s
s
in
g
u
n
it.
Fig
u
r
e
1
2
.
T
im
e
co
m
p
ar
is
o
n
f
o
r
f
o
u
r
ed
it
d
is
tan
ce
in
v
ar
io
u
s
n
u
m
b
e
r
o
f
w
o
r
d
s
Evaluation Warning : The document was created with Spire.PDF for Python.